[SIESTA-L] Message distribution: Authorization denied

[krishna Mohan Grangadhran Prasanna]

Dear Siesta-List,

 

I have been studying the graphene-organic molecule interactions. I use a 10x10 supercell. For pristine graphene SCF run is really fast (~ 5 min on 32 cpus) for Gamma-kpoint. But When I attached TCNE molecule ((CN)2-C=C-(CN)2), and tried the relaxation (where TCNE is about 5 A above the surface),the SCF iteration is really goes beyond 300, 400 etc. i.e. It took much time to converge the SCF and hence the calculation become quite expensive.

Does anyone has an idea to shorten the SCF cycle/convergence time?

If I change SolutionMethod/OccupationFunction/ElectronicTemperature, will it make some improvement? I have played with many parameters but it seems not quite useful. My main settings are given below:

-------------------------------------------------------------

OccupationFunction       MP

ElectronicTemperature   300 K

SpinPolarized            T

PAO.BasisSize           DZP

PAO.BasisType          split

 

ParallelOverK           .true.       

processorY              4   

blocksize                 8   

 

SolutionMethod        diagon

MD.TypeOfRun         CG

MD.NumCGsteps       200

MD.MaxCGDispl        0.05 Ang

MD.MaxForceTol      0.01 eV/Ang

MaxSCFIterations     500

DM.UseSaveDM       .true.  

MD.UseSaveCG       .true.  

DM.MixSCF1           .true.

DM.NUmberKick       20

 

DM.NumberPulay      20

DM.MixingWeight     0.1

DM.Tolerance        1.d-4  

-------------------------------------------------------------

Hope somebody can help me...

 

Thanks in advance,